Silicon Carbide And Gallium Nitride Bring New Challenges For Semiconductor Test

The shift towards compound semiconductors for high-performance power systems requires new testing approaches.

popularity

In the era of megatrends such as electric vehicles (EVs), new technologies are emerging to keep up with evolving demands. One example of this is the evolution of compound semiconductors that use silicon carbide (SiC) and gallium nitride (GaN) for high-performance power systems.

Innovating test protocols to handle wide bandgap materials

For many power-related applications, the semiconductor industry is moving away from silicon to new ‘wide bandgap’ semiconductor materials like SiC and GaN. Due to their unique material properties, SiC and GaN can handle much higher voltages and currents compared to traditional silicon-based semiconductors and can operate at higher frequencies (see figure 1). They are optimized for power conversion applications, for example, SiC is used for traction inverters and in charging stations for EVs, and GaN for industrial and commercial power chargers.

Fig. 1: This diagram shows the significant advantages of SiC and GaN when compared to silicon (source: www.researchgate.net).

Successful testing of these devices requires specialized equipment that can safely and accurately measure very high voltage and current levels; this includes managing the overall increased risk associated with high-power test environments.

SiC and GaN devices also have superior thermal conductivity (see also figure 1), allowing them to operate at higher temperatures. This property, while advantageous in performance, introduces new challenges in thermal testing. Test equipment must be able to simulate and measure device performance under high temperature conditions. Further, maintaining consistent and controlled test environments is crucial to avoid damage and ensure reliable performance data.

SiC and GaN devices are often used in high-frequency applications due to their fast switching capabilities. Testing for dynamic performance, such as switching speed, efficiency, and electromagnetic interference (EMI), is more complex than for silicon devices. Accurate characterization of these parameters requires advanced test methodologies and equipment capable of capturing high-speed events and the fine details of the device’s switching behavior.

The crystalline structures of SiC are also more prone to defects than silicon, which can impact device reliability and performance. Test processes must include thorough reliability testing over extended periods and under varying environmental conditions. It is therefore even more crucial to test SiC devices to weed out infant failures on die even before singulation, and certainly on individual die after singulation. This provides customers with “known good die” (KGD) that can then be packaged as discrete devices or put in parallel with other die into a module. If KGD testing is not done, then a single bad – or low performing – die can cause the entire module to fail, which is a very expensive proposition for power module makers.

Fig. 2: Defects are typically abundant in SiC substrate formation. By contrast, silicon wafers are pristine and defect-free single-crystal substrates (source: Chen, PC., Miao, WC., Ahmed, T. et al. Defect Inspection Techniques in SiC. Nanoscale Res Lett 17, 30 (2022). https://doi.org/10.1186/s11671-022-03672-w).

Teradyne has extended the capabilities of the ETS-88 platform to enable testing of SiC and GaN devices. This includes static DC testing as well as high-speed switching AC testing.

Tapping into AI for optimal data and test efficiency

The overall complexity of testing SiC and GaN devices, particularly in predicting probe wear and managing high-power testing, drives the need for AI and machine learning integration in test systems. For example, with silicon carbide, testing requires the use of thousands of volts and thousands of amps and AI is being used to predict the reliability of the needles that touch a wafer. This empowers better insight, for instance, letting the fab manufacturing team know the needles are going to wear out in a few minutes, hours, or days and to have inventory ready to replace them.

To address this need, Teradyne has been embracing AI and machine learning integration via the Archimedes analytics platform to help optimize test conditions, predict failures, and improve the overall efficiency of the testing process. This allows for immediate adjustments during the test process, improving accuracy and boosting yield. It’s clear that the demand for SiC and GaN test solutions is driving innovation. This includes not only advanced testers but also software tools and diagnostic systems tailored to the unique properties of wide bandgap semiconductors.

Responding with a proactive, collaborative approach to the future

Materials like silicon carbide and gallium nitride are pushing the limits of conventional test methods, requiring innovative solutions to ensure reliability and efficiency in the test process. Teradyne’s ETS-88 is well-suited for these devices, with high throughput and low cost of test for a number of applications.

Continued innovation and preparation for a fast-evolving semiconductor market are critical, allowing ATE to play a crucial role in meeting manufacturers’ needs in a complex and demanding development environment.



Leave a Reply


(Note: This name will be displayed publicly)